update deployment create API

Author: VipulMascarenhas

ads/aqua/extension/deployment_handler.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python
-# -*- coding: utf-8 -*-
 # Copyright (c) 2024 Oracle and/or its affiliates.
 # Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
 
@@ -8,8 +7,8 @@
 from tornado.web import HTTPError
 
 from ads.aqua.common.decorator import handle_exceptions
-from ads.aqua.extension.errors import Errors
 from ads.aqua.extension.base_handler import AquaAPIhandler
+from ads.aqua.extension.errors import Errors
 from ads.aqua.modeldeployment import AquaDeploymentApp, MDInferenceResponse
 from ads.aqua.modeldeployment.entities import ModelParams
 from ads.config import COMPARTMENT_OCID, PROJECT_OCID
@@ -66,8 +65,8 @@ def post(self, *args, **kwargs):
         """
         try:
             input_data = self.get_json_body()
-        except Exception:
-            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT)
+        except Exception as ex:
+            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT) from ex
 
         if not input_data:
             raise HTTPError(400, Errors.NO_INPUT_DATA)
@@ -100,6 +99,8 @@ def post(self, *args, **kwargs):
         health_check_port = input_data.get("health_check_port")
         env_var = input_data.get("env_var")
         container_family = input_data.get("container_family")
+        ocpus = input_data.get("ocpus")
+        memory_in_gbs = input_data.get("memory_in_gbs")
 
         self.finish(
             AquaDeploymentApp().create(
@@ -119,6 +120,8 @@ def post(self, *args, **kwargs):
                 health_check_port=health_check_port,
                 env_var=env_var,
                 container_family=container_family,
+                ocpus=ocpus,
+                memory_in_gbs=memory_in_gbs,
             )
         )
 
@@ -153,9 +156,7 @@ def validate_predict_url(endpoint):
                 return False
             if not url.netloc:
                 return False
-            if not url.path.endswith("/predict"):
-                return False
-            return True
+            return url.path.endswith("/predict")
         except Exception:
             return False
 
@@ -170,8 +171,8 @@ def post(self, *args, **kwargs):
         """
         try:
             input_data = self.get_json_body()
-        except Exception:
-            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT)
+        except Exception as ex:
+            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT) from ex
 
         if not input_data:
             raise HTTPError(400, Errors.NO_INPUT_DATA)
@@ -192,10 +193,10 @@ def post(self, *args, **kwargs):
         )
         try:
             model_params_obj = ModelParams(**model_params)
-        except:
+        except Exception as ex:
             raise HTTPError(
                 400, Errors.INVALID_INPUT_DATA_FORMAT.format("model_params")
-            )
+            ) from ex
 
         return self.finish(
             MDInferenceResponse(prompt, model_params_obj).get_model_deployment_response(
@@ -236,8 +237,8 @@ def post(self, *args, **kwargs):
         """
         try:
             input_data = self.get_json_body()
-        except Exception:
-            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT)
+        except Exception as ex:
+            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT) from ex
 
         if not input_data:
             raise HTTPError(400, Errors.NO_INPUT_DATA)

ads/aqua/modeldeployment/deployment.py

@@ -3,7 +3,7 @@
 # Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
 
 import logging
-from typing import Dict, List, Union
+from typing import Dict, List, Optional, Union
 
 from ads.aqua.app import AquaApp, logger
 from ads.aqua.common.enums import (
@@ -102,6 +102,8 @@ def create(
         health_check_port: int = None,
         env_var: Dict = None,
         container_family: str = None,
+        memory_in_gbs: Optional[float] = None,
+        ocpus: Optional[float] = None,
     ) -> "AquaDeployment":
         """
         Creates a new Aqua deployment
@@ -142,6 +144,10 @@ def create(
             Environment variable for the deployment, by default None.
         container_family: str
             The image family of model deployment container runtime. Required for unverified Aqua models.
+        memory_in_gbs: float
+            The memory in gbs for the shape selected.
+        ocpus: float
+            The ocpu count for the shape selected.
         Returns
         -------
         AquaDeployment
@@ -325,6 +331,11 @@ def create(
                 log_id=predict_log_id,
             )
         )
+        if memory_in_gbs and ocpus and infrastructure.shape_name.endswith("Flex"):
+            infrastructure.with_shape_config_details(
+                ocpus=ocpus,
+                memory_in_gbs=memory_in_gbs,
+            )
         # configure model deployment runtime
         container_runtime = (
             ModelDeploymentContainerRuntime()
@@ -338,6 +349,7 @@ def create(
             .with_overwrite_existing_artifact(True)
             .with_remove_existing_artifact(True)
         )
+
         # configure model deployment and deploy model on container runtime
         deployment = (
             ModelDeployment()

tests/unitary/with_extras/aqua/test_data/deployment/aqua_create_gguf_deployment.yaml

@@ -0,0 +1,40 @@
+kind: deployment
+spec:
+  createdBy: ocid1.user.oc1..<OCID>
+  displayName: model-deployment-name
+  freeformTags:
+    OCI_AQUA: active
+    aqua_model_name: model-name
+  id: "ocid1.datasciencemodeldeployment.oc1.<region>.<MD_OCID>"
+  infrastructure:
+    kind: infrastructure
+    spec:
+      bandwidthMbps: 10
+      compartmentId: ocid1.compartment.oc1..<OCID>
+      deploymentType: SINGLE_MODEL
+      policyType: FIXED_SIZE
+      projectId: ocid1.datascienceproject.oc1.iad.<OCID>
+      replica: 1
+      shapeName: "VM.Standard.A1.Flex"
+      shapeConfigDetails:
+        memoryInGBs: 60.0
+        ocpus: 10
+    type: datascienceModelDeployment
+  lifecycleState: CREATING
+  modelDeploymentUrl: "https://modeldeployment.customer-oci.com/ocid1.datasciencemodeldeployment.oc1.<region>.<MD_OCID>"
+  runtime:
+    kind: runtime
+    spec:
+      env:
+        BASE_MODEL: service_models/model-name/artifact
+        BASE_MODEL_FILE: model-name.gguf
+        MODEL_DEPLOY_ENABLE_STREAMING: 'true'
+        MODEL_DEPLOY_PREDICT_ENDPOINT: /v1/completions
+        MODEL_DEPLOY_HEALTH_ENDPOINT: /v1/models
+      healthCheckPort: 8080
+      image: "dsmc://image-name:1.0.0.0"
+      modelUri: "ocid1.datasciencemodeldeployment.oc1.<region>.<MODEL_OCID>"
+      serverPort: 8080
+    type: container
+  timeCreated: 2024-01-01T00:00:00.000000+00:00
+type: modelDeployment

tests/unitary/with_extras/aqua/test_data/deployment/deployment_config.json

@@ -5,9 +5,26 @@
         "TGI_PARAMS": "--max-stop-sequences 6",
         "VLLM_PARAMS": "--max-model-len 4096"
       }
+    },
+    "VM.Standard.A1.Flex": {
+      "parameters": {},
+      "shape_info": {
+        "configs": [
+          {
+            "memory_in_gbs": 128,
+            "ocpu": 32
+          },
+          {
+            "memory_in_gbs": 256,
+            "ocpu": 64
+          }
+        ],
+        "type": "CPU"
+      }
     }
   },
   "shape": [
-    "VM.GPU.A10.1"
+    "VM.GPU.A10.1",
+    "VM.Standard.A1.Flex"
   ]
 }

tests/unitary/with_extras/aqua/test_data/ui/container_index.json

@@ -1,5 +1,27 @@
 {
   "containerSpec": {
+    "odsc-llama-cpp-serving": {
+      "cliParam": "",
+      "envVars": [
+        {
+          "MODEL_DEPLOY_PREDICT_ENDPOINT": "/v1/completions"
+        },
+        {
+          "MODEL_DEPLOY_HEALTH_ENDPOINT": "/v1/models"
+        },
+        {
+          "MODEL_DEPLOY_ENABLE_STREAMING": "true"
+        },
+        {
+          "PORT": "8080"
+        },
+        {
+          "HEALTH_CHECK_PORT": "8080"
+        }
+      ],
+      "healthCheckPort": "8080",
+      "serverPort": "8080"
+    },
     "odsc-tgi-serving": {
       "cliParam": "--sharded true --trust-remote-code",
       "envVars": [
@@ -39,6 +61,14 @@
       "serverPort": "8080"
     }
   },
+  "odsc-llama-cpp-serving": [
+    {
+      "displayName": "llama.cpp",
+      "name": "dsmc://odsc-llama-cpp-serving",
+      "type": "inference",
+      "version": "0.2.75.2"
+    }
+  ],
   "odsc-llm-evaluate": [
     {
       "name": "dsmc://odsc-llm-evaluate",